The benefits of event: driven energy accounting in power-sensitive systems

@article{Bellosa2000TheBO,
  title={The benefits of event: driven energy accounting in power-sensitive systems},
  author={Frank Bellosa},
  journal={Proceedings of the 9th workshop on ACM SIGOPS European workshop: beyond the PC: new challenges for the operating system},
  year={2000}
}
  • Frank Bellosa
  • Published 17 September 2000
  • Computer Science
  • Proceedings of the 9th workshop on ACM SIGOPS European workshop: beyond the PC: new challenges for the operating system
A prerequisite of energy-aware scheduling is precise knowledge of any activity inside the computer system. Embedded hardware monitors (e.g., processor performance counters) have proved to offer valuable information in the field of performance analysis. The same approach can be applied to investigate the energy usage patterns of individual threads. We use information about active hardware units (e.g., integer/floating-point unit, cache/memory interface) gathered by event counters to establish a… 

Figures from this paper

Process cruise control: event-driven clock scaling for dynamic power management
TLDR
An energy-aware scheduling policy for non-real-time operating systems that benefits from event counters is proposed and energy measurements of the target architecture under variable load show the advantage of the proposed approach.
E MPYA : An Energy-Aware Middleware Platform for Dynamic Applications
TLDR
EMPYA is presented, an energy-aware programming and execution platform that frees application programmers from the need to take care of energy efficiency and combines techniques from different software and hardware levels to effectively and efficiently minimize the resource footprint of an application during periods of low utilization.
A study of hardware performance monitoring counter selection in power modeling of computing systems
  • R. Zamani, A. Afsahi
  • Computer Science
    2012 International Green Computing Conference (IGCC)
  • 2012
TLDR
This paper presents a method to select the most correlated set of PMC events for a given application and shows that power estimates can be enhanced more than common regression methods.
Performance Events Based Full System Estimation on Application Power Consumption
  • Shu Yang, Zhongzhi Luan, Binyang Li, Ge Zhang, Tianming Huang, D. Qian
  • Computer Science, Engineering
    2016 IEEE 18th International Conference on High Performance Computing and Communications; IEEE 14th International Conference on Smart City; IEEE 2nd International Conference on Data Science and Systems (HPCC/SmartCity/DSS)
  • 2016
TLDR
The power models proposed in this paper can provide software developers or system designers with visible power behaviors for the applications and can make decisions or change code to develop more energy-efficient software.
EnergyBudgets: Integrating Physical Energy Measurement Devices into Systems Software
TLDR
A modular analysis approach, EnergyBudgets, which bridges external energy measurement hardware to the Linux perf subsystem and shows that energy budgets accurately measure the energy consumed by different workloads and allow for an overhead-reduction on the SUT by 20% to 51% in comparison to regular timers, while still guaranteeing the same level of precision.
Techniques for Real-System Characterization of Java Virtual Machine Energy and Power Behavior
TLDR
This paper presents techniques for characterizing Java power and performance, as well as results from applying these techniques to the Jikes RVM, for some of the most salient Java Virtual Machine components.
Dynamic Power Estimation with Hardware Performance Counters Support on Multi-core Platform
TLDR
By analyzing the hardware events in processor through performance counters, a unified run-time power estimation model is introduced and can serve as a foundation for intelligent, power-aware systems that can dynamically balance power assignment and smooth peak power at run- time.
HaPPy: Hyperthread-aware Power Profiling Dynamically
TLDR
A hyperthread-aware power model that differentiates between the states when both hardware threads of a core are in use, and when only one thread is in use is introduced, able to accurately attribute power to each logical CPU in modern servers.
Next Generation Hardware Monitoring Infrastructure for Multi-core Resource Auditing
TLDR
This work studies the design and implementation of next generation hardware monitoring infrastructure, which offers a collection of hardware counters for auditing architectural events on hardware units and can be used by programmers and performance analysts for audited performance bottlenecks and consequently, optimizing application performance.
Accurate on-line prediction of processor and memoryenergy usage under voltage scaling
TLDR
A systematic approach to building platform-specific but workload-independent performance and power models that can accurately predict the energy use of a wide class of applications and is highly responsive to changes in the application behaviour.
...
...

References

SHOWING 1-10 OF 34 REFERENCES
Continuous profiling: where have all the cycles gone?
TLDR
The Digital Continuous Profiling Infrastructure is a sampling-based profiling system designed to run continuously on production systems, supporting multiprocessors, works on unmodified executables, and collects profiles for entire systems, including user programs, shared libraries, and the operating system kernel.
Continuous profiling: where have all the cycles gone?
TLDR
The DIGITAL Continuous Profiling Infrastructure is described, a sampling-based profiling system designed to run continuously on production systems, and guides users and automated optimizers to the causes of performance problems and provides important insights for fixing them.
The Performance Implications of Locality Information Usage in Shared-Memory . . .
  • Computer Science
  • 1996
TLDR
A prototype implementation shows that a locality-conscious scheduler outperforms approaches ignoring locality information, and proposes novel scheduling policies based on locality information derived from cache miss counters.
The Performance Limits of Locality Information Usage in Shared-Memory Multiprocessors
TLDR
A prototype implementation shows that a locality-conscious scheduler outperforms approaches ignoring locality information, and proposes novel scheduling policies based on locality information derived from cache miss counters.
Monitoring system activity for OS-directed dynamic power management
TLDR
This work used its monitoring tool to collect data on the usage of system resources for a laptop computer, under several workload conditions, and shows that resource usage is strongly resource and workload dependent, and that on-line usage monitoring capability is a critical issue of the implementation of effective power management policies.
System Support for Energy Management in Mobile and Embedded Workloads: A White Paper
TLDR
This research discusses the vulnerability of one such application to power consumption and examines the need to reduce the energy consumed in using these devices, thereby extending the lifetime of the batteries that power them as well as reducing the impact of the heat and noise produced by their operation.
Scheduling for reduced CPU energy
TLDR
A new metric for cpu energy performance, millions-of-instructions-per-joule (MIPJ), and several methods for varying the clock speed dynamically under control of the operating system, and examine the performance of these methods against workstation traces.
Energy-efficient design of battery-powered embedded systems
  • T. Simunic, L. Benini, G. Micheli
  • Computer Science, Engineering
    Proceedings. 1999 International Symposium on Low Power Electronics and Design (Cat. No.99TH8477)
  • 1999
Energy-efficient design of battery-powered embedded systems demands optimizations in both hardware and software. In this work we leverage cycle-accurate energy consumption models to explore compiler
Performance counters and state sharing annotations: a unified approach to thread locality
TLDR
A combined approach for improving thread locality that uses the bardware performance monitors of modem processors and program-centric code annotations to guide thread scheduling on SMPs and relies on a shared state cache model to compute expected thread footprints in the cache on-line.
FOLLOW-ON SCHEDULING USING TLB INFORMATION TO REDUCE CACHE MISSES
TLDR
This approach to improve system performance uses information derived from the translation lookaside buffer (TLB) to detect kernel threads which share lots of memory pages which results in a high cache reusage, in few cache misses, and in a good system performance.
...
...